Information retrieval is the process of fulfilling a user?s need for information by locating items in a data collection that are similar to a complex query that is often posed in natural language. Latent Semantic Indexing (LSI) was the predominant technique employed at the National Institute of Standards and Technology?s Text Retrieval Conference for many years until limitations of its scalability to large data sets were discovered. This thesis describes SCRIBE, a modification of LSI with improved scalability. SCRIBE clusters its semantic index into discrete volumes described by high-dimensional extensions to computer graphics data structures. SCRIBE?s clustering strategy limits the number of items that must be searched and provides for sub-linear time complexity in the number of documents. Experimental results with a large, natural language document collection demonstrate that SCRIBE achieves retrieval accuracy similar to LSI but requires 1/10 the time.

The role of factors previously implicated as leading to confirmation bias during hypothesis testing was explored. Confirmation bias is a phenomenon in which people select cases for testing when the expected results of the case are more likely to support their current belief than falsify it. Klayman (1995) proposed three primary determinants for confirmation bias. Klayman and his colleagues proposed that a general positive testing strategy leads to the phenomenon of confirmation bias. According to Klayman’s account, participants in previous research were not actively working to support their hypothesis. Rather, they were applying a valid hypothesis testing strategy that works well outside of laboratory tasks. In laboratory tasks, such as Wason’s 2-4-6 task (Wason, 1960), the strategy failed because the nature of the task takes advantage of particular flaws in the positive testing behavior participants learned through their experience with the real-world. Given Klayman’s proposed set of determinants for the positive testing strategy phenomenon, treatments were developed that would directly violate the assumptions supporting application of the positive testing strategy. If participants were able to identify and act on these violations of the assumptions, the number of positive tests was expected to be reduced. The test selection portion of the Mynatt, Doherty, and Tweney (1977) microworld experiment was modified with additional instruction conditions and a new scenario description to investigate the impact of the treatments to reduce confirmation bias in test selection. Despite expectations, the thematic content modifications and determinant-targeting instruction conditions had no effect on participant positive test selection.

No description available.

No description available.

No description available.

Breast cancer (BC) is the second most diagnosed cancer in Canadian women. Early detection of this cancer is critical to improve patient survival and prognoses. Exosomes are proposed to be involved in tumor proliferation through the transfer of diverse biomolecules, including metabolites. The use of exosomes as biomarkers for early diagnosis of BC has recently garnered interest due to them having unique biomolecules in diseased cohorts. Hence, an untargeted metabolomic analysis of BC exosomes was performed using nano high-performance liquid chromatography coupled to tandem mass spectrometry (nLC-MS/MS) for BC diagnostic biomarker discovery. A total of 9 independent metabolite samples from non-tumorogenic MCF10A and highly metastatic MDA-MB-231 cell lines were analyzed. Bioinformatic analysis revealed 27 potential metabolite candidates unique to MDA-MB-231. Amongst 4 metabolites tested, one, N-Acetyl-L-Phenylalanine, was successfully validated. Overall, this study reveals that exosomes possess metabolites that can be candidates for early BC diagnosis.

This research is focused on identifying a single-step procedure that, upon adapting to the data through estimating the unknown parameters, would asymptotically control the False Discovery Rate when testing a large number of hypotheses simultaneously, and exploring some of the characteristics of this procedure. / Statistics

<p dir="ltr">In computer-aided drug discovery, methods that are approximate, but computationally inexpensive play an essential role during the initial phase of the discovery process. Although often inaccurate, they enable the screening of vast drug libraries to identify potential inhibitors with favorable activities, before large amounts of computational resources could be dedicated to studying these individual molecules. This thesis presents<b> </b>such an approach, based on the concept of hydrogen bond wrapping, to study protein-ligand interactions in the context of drug discovery. The ‘wrapping’ refers to the tendency of hydrophobic groups to surround a hydrogen bond in water, leading to its desolvation, thereby stabilizing it.</p><p dir="ltr">Herein, a molecular descriptor was employed, which quantifies the extent of hydrophobic wrapping around a protein’s backbone hydrogen bonds (BHBs) and could help speed up the discovery process by providing cues for the design or optimization of inhibitors. Additionally, these insights could help tailor not just the binding affinity of inhibitors, but also their specificity toward an intended target protein. The human transmembrane protease serine 2 (TMPRSS2) was used as an illustrative target protein due to the pressing need for COVID-19 therapeutics, and since the current understanding of the binding mechanisms of known TMPRSS2 inhibitors is limited.</p><p dir="ltr">Molecular docking with a Generalized Born - surface area (GBSA) scoring function was first performed to virtually screen for TMPRSS2 inhibitors. The molecular descriptor was then used to analyze the change in wrapping groups of TMPRSS2 BHBs due to docked ligands, with the aim of identifying BHBs with a high propensity for desolvation. The BHBs involving residues Cys437, Gln438, Asp440, and Ser441 of TMPRSS2 were seen to have some of the largest average increases in wrapping. These general results were also compared to results from docking of the known TMPRSS2 inhibitors, camostat, and nafamostat.</p><p dir="ltr">The data generated from docking were then used to examine potential applications of the wrapping molecular descriptor using machine learning techniques: (i) for prediction of the solvent-accessible surface area term ΔG_sa of the GBSA score using regression and (ii) for classifying the solvent interactions of a TMPRSS2-inhibitor complex as favorable or unfavorable. The descriptor was seen to be only weakly related to ΔG_sa; the best-performing regression model had a Pearson correlation coefficient of 0.76 between the predictions and the actual values. The ability of the descriptor to classify solvent interactions was more satisfactory, with a highest value for area under the receiver operating characteristic curve of 0.75.</p><p dir="ltr">The descriptor was then used to analyze the effect of inhibitor binding on the dynamics of TMPRSS2 BHBs. For this, molecular dynamics simulation was carried out for the uncomplexed TMPRSS2, as well as its complex with known inhibitors and hit molecules from docking. The binding of these ligands was seen to improve the stability of TMPRSS2; certain BHBs which were unstable or not formed in the uncomplexed case, showed increased stability. These prominently included a couple of BHBs identified from docking as having gained a large increase in wrapping. The improved stability coincided with an increase in wrapping groups in several cases. The descriptor also successfully rationalized the desolvation of a few BHBs due to inhibitor binding.</p><p dir="ltr">This work demonstrates the potential application of the concept of hydrogen bond wrapping in understanding the mechanism of inhibitor binding and the resultant desolvation effects on a protein’s BHBs, without computationally expensive calculations. While the analysis methods require further improvement, the wrapping descriptor shows promising results and could be developed into a simple, yet powerful tool for drug discovery.</p>

Enterprise decision making is continuously transforming in the wake of ever increasing amounts of data. Organizations are collecting massive amounts of data in their quest for knowledge nuggets in form of novel, interesting, understandable patterns that underlie these data. The search for knowledge is a multi-step process comprising of various phases including development of domain (business) understanding, data understanding, data preparation, modeling, evaluation and ultimately, the deployment of the discovered knowledge. These phases are represented in form of Knowledge Discovery and Data Mining (KDDM) Process Models that are meant to provide explicit support towards execution of the complex and iterative knowledge discovery process. Review of existing KDDM process models reveals that they have certain limitations (fragmented design, only a checklist-type description of tasks, lack of support towards execution of tasks, especially those of the business understanding phase etc) which are likely to affect the efficiency and effectiveness with which KDDM projects are currently carried out. This dissertation addresses the various identified limitations of existing KDDM process models through an improved model (named the Integrated Knowledge Discovery and Data Mining Process Model) which presents an integrated view of the KDDM process and provides explicit support towards execution of each one of the tasks outlined in the model. We also evaluate the effectiveness and efficiency offered by the IKDDM model against CRISP-DM, a leading KDDM process model, in aiding data mining users to execute various tasks of the KDDM process. Results of statistical tests indicate that the IKDDM model outperforms the CRISP model in terms of efficiency and effectiveness; the IKDDM model also outperforms CRISP in terms of quality of the process model itself.

No description available.